Re: lynx-dev strcasecomp/strncasecomp

[pg]

In a recent note, T.E.Dickey said:

> Date: Fri, 16 Jul 1999 20:40:31 -0400 (EDT)
> 
> there's three versions - M$, ASCII and EBCDIC.  (or that's my impression).
> Does Hiroyuki Senshu's (or is that Hiroyuki's) patch break the EBCDIC version?
> 
I've looked more closely.

ASCII and EBCDIC go through a common path, both invoking TOLOWER (which
may split for ASCII and EBCDIC -- I haven't checked.)

The new code doesn't replace the prior; it's wrapped in "#ifdef SH_EX".
I haven't tried it, but it seems that with SH_EX defined, I'd get a
multiply defined function error.

> (I figured people would argue about whether the speed is really that 
> different,
> and we'd get some sort of consensus about whether it was enough to be useful).
>   
Was speed the objective?  I notice that some of the new code exploits the
"register" storage class.  Author's whim, I guess.

-- gil 
-- 
StorageTek
INFORMATION made POWERFUL